[Mono] Fix uninitialized vtable bug#67746
Conversation
simonrozsival
requested review from
lambdageek,
thaystg and
vargaz
as code owners
ghost
added
the
|
Tagging subscribers to this area: @dotnet/ncl Issue DetailsFixes #67402
|
|
/azp run runtime-extra-platforms |
|
Azure Pipelines successfully started running 1 pipeline(s). |
eerhardt
left a comment
eerhardt
left a comment
There was a problem hiding this comment.
Looks good to me. We may want to consider a write test, but given the code quickly calls into a common method I don't think it is crucial.
|
/backport to release/6.0 |
|
Started backporting to release/6.0: https://github.com/dotnet/runtime/actions/runs/2115720964 |
…tem-io-stream-vtable-crash
|
/azp run runtime-extra-platforms |
|
Azure Pipelines successfully started running 1 pipeline(s). |
|
Thanks so much for fixing this. Is there a workaround that doesn't include disabling AOT? |
|
The |
|
@jahmai-ca right now I don't know any other workaround |
This reverts commit daf631e.
|
/azp run runtime-extra-platforms |
|
Azure Pipelines successfully started running 1 pipeline(s). |
|
/azp run runtime-extra-platforms |
|
Azure Pipelines successfully started running 1 pipeline(s). |
|
/azp run runtime-extra-platforms |
|
Azure Pipelines successfully started running 1 pipeline(s). |
|
/azp run runtime-extra-platforms |
|
Azure Pipelines successfully started running 1 pipeline(s). |
|
/azp run runtime-extra-platforms |
|
Azure Pipelines successfully started running 1 pipeline(s). |
|
/azp run runtime-extra-platforms |
|
Azure Pipelines successfully started running 1 pipeline(s). |
…ozsival/runtime into fix-system-io-stream-vtable-crash
|
/azp run runtime-extra-platforms |
|
Azure Pipelines successfully started running 1 pipeline(s). |
|
/azp run runtime-extra-platforms |
|
Azure Pipelines successfully started running 1 pipeline(s). |
7 participants
Fixes #67402
The code which checks whether a class derived from
System.IO.Streamhas overridden certain methods didn't explicitly setup the vtable before it tried to dereference the vtable pointer. When AOT was enabled the vtable pointer was null and dereferencing it caused a crash.I also checked the other places where we access vtable and in most cases, we call themono_class_setup_vtablemethod before we access the vtable except in two cases where it isn't obvious that the vtable is setup:cominterop.c:502mini-trampolines.c:593I don't want to add any modifications to those pieces of code in this PR but I'll check those two instances in more detail and I might follow-up with another PR later.EDIT: I analzyed the two pieces of code I mentioned previously and I came to the conclusion that accessing the vtable won't be a problem in those cases.